Imperfect Form

Demo

About

The problem it solves Imperfect Form tackles major challenges in fitness like obesity, injury prevention, and workout motivation by turning exercise into a transparent, gamified, and engaging experience. Real-time pose detection helps users improve form and avoid injuries, while blockchain leaderboards and social integration add accountability and competition that make workouts less boring and more rewarding. By combining verifiable human participation with gamification, it encourages consistent activity, healthier habits, and a sense of community — reducing dropout rates and helping users stick to long-term fitness goals. Smart Contract Address(es) - **Base Mainnet**: │ │ `0x60228F4f4F1A71e9b43ebA8C5A7ecaA7e4d4950B`

How it's made

Challenges we ran into One of the biggest challenges in building my onchain fitness app has been elevating the UI/UX to a truly high-quality standard. Since fitness tracking is inherently user-facing and interactive, every detail matters — from how smooth the interface feels to how intuitively users can navigate the app. Iterating on design and refining the flow took significant time, but these improvements were essential to ensure that users feel motivated and engaged when tracking their workouts. Another key hurdle was enabling seamless participation with a variety of wallet options. Because users come with different levels of Web3 familiarity, it was important to provide flexible onramps without sacrificing security or accessibility. Integrating multiple wallet solutions required careful handling of edge cases, testing across environments, and refining the onboarding flow so that both crypto-native and first-time users could engage without friction. I still have work to do here!! Finally, maintaining and improving the performance of external models, particularly those built with TensorFlow, presented a technical challenge. These models power the app’s core features, so ensuring they run efficiently across devices while staying accurate was critical. To address this, I optimized model loading, streamlined inference calls, and implemented performance monitoring tools — all of which help balance computational demands with a responsive user experience. I still have work to do here!!